Abstract

We extend the concept of the reverse distance transform for binary images to a fuzzy framework, leading to the reversefuzzy distance transform. Furthermore, we introduce centres of maximal fuzzy balls intended for, e.g. fuzzy distance based skeletonization of a fuzzy object. Finally, we propose to use the reverse fuzzy distance transform as an alternative region growing process to the seeded watershed segmentation, in order to further emphasize the shape of fuzzy objects. We make a comparison using an application where the aim is to identify the cytoplasms in fluorescence microscopy images of cells. (C) 2008 Elsevier B.V. All rights reserved.

Keywords

fuzzy distance; region growing; watershed segmentation

Published in

Pattern Recognition Letters
2008, volume: 29, number: 7, pages: 888-896
Publisher: ELSEVIER SCIENCE BV

Authors' information